Systemic risk detection using an entropy approach in portfolio selection strategy

Abstract

This paper focuses on the investigation and detection of systemic risk. Such risk significantly affects the financial markets and the banking sector, and is fundamental for macro-prudential regulation. To address this issue, we propose an early warning system to anticipate periods of distress. In particular, we consider systemic risk from the investors' perspective, developing optimal portfolio strategies that incorporate such an early warning system based on different entropy measures to predict and hedge the occurrence of systemic risk. On top of this, we introduce a rule that, in periods of crisis, triggers a switch to a risk-free portfolio. In order to determine the optimal composition of a portfolio, we use a new double-optimization strategy, which consists of the maximization of selected performance ratios in the first step and the minimization of selected systemic risk indicators (CoVaR, Marginal expected shortfall) for a given expected return in the second step. An empirical analysis shows that the proposed strategy allows reducing the total risk of the portfolio and generally improves its profitability. We finally discuss how the introduction of these investment strategies may affect the overall stability of the financial system.